Effect of blood pool 18F-FDG activity on liver SUV: a calibration methodology study

Abstract

Objective To identify the effect of blood pool 18F-FDG activity on liver SUV and to investigate the optimal normalization method. Methods PET/CT and common serological examination items from 1 018 subjects were retrospectively collected. Mean SUV of liver and blood were recorded as SUVmean(L) and SUVmean(B), respectively. The difference and quotient of SUVmean(L) and SUVmean(B) were calculated as SUVmean(L-B) and SUVmean(L/B), respectively. CV of SUVmean(L), SUVmean(L-B) and SUVmean(L/B) were calculated to assess their inter-individual variations. Pearson correlation analysis was used to evaluate the relationship of SUVmean(L), SUVmean(L-B), SUVmean(L/B) with SUVmean(B). Multiple linear stepwise regression was performed to identify their vulnerability to common serological examination items. Results CV of SUVmean(L/B) (15.1%) was less than that of SUVmean(L) (23.2%) and SUVmean(L-B) (40.6%). Correlation between SUVmean(L) and SUVmean(B) (r=0.820, P<0.001) was more significant than that between SUVmean(L-B) and SUVmean(B) (r=0.205, P<0.001) as well as between SUVmean(L/B) and SUVmean(B) (r=-0.376, P<0.001). Blood glucose and BMI correlated with SUVmean(L) and SUVmean(L/B), but not with SUVmean(B). Age and HDL correlated with SUVmean(L) and SUVmean(B), but not with SUVmean(L/B). Fatty liver was significantly associated with SUVmean(L/B)(β=-0.047, P<0.001), but not with SUVmean(L) and SUVmean(B). Conclusions 18F-FDG activity of blood pool affects liver SUV. SUVmean(L/B) is a simple and reliable normalization method since its inter-individual variation and vulnerability to common serological examination items are relatively lower than liver SUV. Key words: Liver; Standardized uptake value; Tomography, emission-computed; Tomography, X-ray computed; Deoxyglucose